// ===============================================================================
// = Merge overlapping BED entries into a single entry, accounting for strandedness =
// ================================================================================
void BedMerge::MergeBedStranded() {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
_bed->loadBedFileIntoMapNoBin();
// loop through each chromosome and merge their BED entries
masterBedMapNoBin::const_iterator m = _bed->bedMapNoBin.begin();
masterBedMapNoBin::const_iterator mEnd = _bed->bedMapNoBin.end();
for (; m != mEnd; ++m) {
// bedList is already sorted by start position.
string chrom = m->first;
// make a list of the two strands to merge separately.
vector<string> strands(2);
strands[0] = "+";
strands[1] = "-";
// do two passes, one for each strand.
for (unsigned int s = 0; s < strands.size(); s++) {
int mergeCount = 1;
int numOnStrand = 0;
vector<string> names;
vector<string> scores;
// merge overlapping features for this chromosome.
int start = -1;
int end = -1;
vector<BED>::const_iterator bedItr = m->second.begin();
vector<BED>::const_iterator bedEnd = m->second.end();
for (; bedItr != bedEnd; ++bedItr) {
// if forcing strandedness, move on if the hit
// is not on the current strand.
if (bedItr->strand != strands[s]) { continue; }
else { numOnStrand++; }
if ( (((int) bedItr->start - end) > _maxDistance) ||
(end < 0))
{
if (start >= 0) {
ReportStranded(chrom, start, end, names,
scores, mergeCount, strands[s]);
// reset
mergeCount = 1;
names.clear();
scores.clear();
}
start = bedItr->start;
end = bedItr->end;
if (!bedItr->name.empty()) names.push_back(bedItr->name);
if (!bedItr->score.empty()) scores.push_back(bedItr->score);
}
else {
if ((int) bedItr-> end > end) end = bedItr->end;
mergeCount++;
if (!bedItr->name.empty()) names.push_back(bedItr->name);
if (!bedItr->score.empty()) scores.push_back(bedItr->score);
}
}
if (start >= 0) {
ReportStranded(chrom, start, end, names,
scores, mergeCount, strands[s]);
}
}
}
}
// ==================================================================================
// = Merge overlapping BED entries into a single entry, accounting for strandedness =
// ==================================================================================
void BedMerge::MergeBedStranded() {
// load the "B" bed file into a map so
// that we can easily compare "A" to it for overlaps
_bed->loadBedFileIntoMapNoBin();
// loop through each chromosome and merge their BED entries
masterBedMapNoBin::const_iterator m = _bed->bedMapNoBin.begin();
masterBedMapNoBin::const_iterator mEnd = _bed->bedMapNoBin.end();
for (; m != mEnd; ++m) {
// bedList is already sorted by start position.
vector<BED> bedList = m->second;
// make a list of the two strands to merge separately.
vector<string> strands(2);
strands[0] = "+";
strands[1] = "-";
// do two passes, one for each strand.
for (unsigned int s = 0; s < strands.size(); s++) {
CHRPOS minStart = INT_MAX;
CHRPOS maxEnd = 0;
bool OIP = false; // OIP = Overlap In Progress. Lame, I realize.
int prev = -1;
unsigned int curr = 0;
int mergeCount = 1;
int numOnStrand = 0;
map<string, bool> names;
// loop through the BED entries for this chromosome
// and look for overlaps
for (curr = 0; curr < bedList.size(); ++curr) {
// if forcing strandedness, move on if the hit
// is not on the current strand.
if (bedList[curr].strand != strands[s]) {
continue; // continue force the next iteration of the for loop.
}
else {
numOnStrand++;
}
// make sure prev points to an actual element on the
// current strand
if (prev < 0) {
if (bedList[curr].strand == strands[s]) {
prev = curr;
}
continue;
}
if ( overlaps(bedList[prev].start, bedList[prev].end,
bedList[curr].start, bedList[curr].end) >= _maxDistance) {
OIP = true;
mergeCount++;
minStart = min(bedList[prev].start, min(minStart, bedList[curr].start));
maxEnd = max(bedList[prev].end, max(maxEnd, bedList[curr].end));
names[bedList[prev].name] = true;
names[bedList[curr].name] = true;
}
else if ( overlaps(minStart, maxEnd,
bedList[curr].start, bedList[curr].end) >= _maxDistance) {
mergeCount++;
minStart = min(minStart, bedList[curr].start);
maxEnd = max(maxEnd, bedList[curr].end);
names[bedList[curr].name] = true;
}
else {
// was there an overlap before the current entry broke it?
if (OIP) {
if (_numEntries) {
cout << bedList[prev].chrom << "\t" << minStart << "\t" << maxEnd << "\t" << mergeCount << "\t" << strands[s] << endl;
}
else if (_reportNames) {
cout << bedList[prev].chrom << "\t" << minStart << "\t" << maxEnd << "\t";
ReportMergedNames(names);
cout << "\t" << strands[s] << endl;
}
else {
cout << bedList[prev].chrom << "\t" << minStart << "\t" << maxEnd << "\t" << strands[s] << endl;
}
}
else {
if ((_numEntries) && (numOnStrand > 0)) {
cout << bedList[prev].chrom << "\t" << bedList[prev].start << "\t" << bedList[prev].end << "\t" << mergeCount << "\t" << strands[s] << endl;
}
else if (_reportNames) {
cout << bedList[prev].chrom << "\t" << bedList[prev].start << "\t" << bedList[prev].end << "\t" << bedList[prev].name << "\t" << strands[s] << endl;
}
else if (numOnStrand > 0) {
cout << bedList[prev].chrom << "\t" << bedList[prev].start << "\t" << bedList[prev].end << "\t" << strands[s] << endl;
}
}
// reset things for the next overlapping "block"
OIP = false;
mergeCount = 1;
minStart = INT_MAX;
maxEnd = 0;
names.clear();
// add the name of the current element in prep for the next block
names[bedList[curr].name] = true;
}
prev = curr;
}
// clean up based on the last entry for the current chromosome
if (OIP) {
if (_numEntries) {
cout << bedList[prev].chrom << "\t" << minStart << "\t" << maxEnd << "\t" << mergeCount << "\t" << strands[s] << endl;
}
else if (_reportNames) {
cout << bedList[prev].chrom << "\t" << minStart << "\t" << maxEnd << "\t";
ReportMergedNames(names);
cout << "\t" << strands[s] << endl;
}
else {
cout << bedList[prev].chrom << "\t" << minStart << "\t" << maxEnd << "\t" << strands[s] << endl;
}
}
else {
if ((_numEntries) && (numOnStrand > 0)) {
cout << bedList[prev].chrom << "\t" << bedList[prev].start << "\t" << bedList[prev].end << "\t" << mergeCount << "\t" << strands[s] << endl;
}
else if ((_reportNames) && (numOnStrand > 0)) {
cout << bedList[prev].chrom << "\t" << bedList[prev].start << "\t" << bedList[prev].end << "\t" << bedList[prev].name << "\t" << strands[s] << endl;
}
else if (numOnStrand > 0) {
cout << bedList[prev].chrom << "\t" << bedList[prev].start << "\t" << bedList[prev].end << "\t" << strands[s] << endl;
}
}
}
}
}
